| The aircraft engine is the core component of the aircraft,and its precise installation is of great significance to the flight safety of the aircraft.In order to change the current situation of low efficiency and high labor intensity of aero-engine installation,based on the comparative study of multi-feature extraction algorithm and SSD(Single Shot Multibox Detector)algorithm,an improved algorithm based on R-D SSD(Res Net-Dilated SSD)is proposed.The installation station detection method realizes the accurate and rapid detection of the installation station of the aircraft engine during the automatic installation process;a centering-fixing axis algorithm based on binocular stereo vision is proposed to realize the transportation of the aircraft engine during the automatic installation process accurate and fast positioning of the installation vehicle.By extracting the gray-scale feature,area feature and directional gradient histogram feature of the installation station,the method of combining morphological analysis and support vector machine classification is used to realize the installation station detection based on multi-feature extraction.In view of the characteristics of the data set containing installation stations of different sizes,based on the idea of multi-scale prediction in the convolutional neural network,through the training and testing of the network model,the installation station detection based on the SSD algorithm is realized.By comparing the detection effects of the two methods,the SSD algorithm is selected as the basic algorithm for research on aero-engine installation station detection for optimization and improvement.In order to solve the defect that the SSD algorithm has low detection accuracy for small target installation stations,an installation station detection method based on the improved R-D SSD algorithm is proposed.R-D SSD integrates the replacement of the backbone network,the increase of the number of preselected boxes for the output feature map,and the introduction of an expansion factor.The measures can effectively solve the problem that the SSD algorithm is not sensitive to the detection of small target installation stations,and meet the detection accuracy requirements and detection speed requirements of the installation stations during the automatic installation of aircraft engine.In order to realize the accurate and rapid positioning of the transport installation vehicle during the automatic installation of aero-engines,based on the installation station detection results based on the R-D SSD algorithm,a centering-fixing axis algorithm based on binocular stereo vision is proposed.Use the Hough transform round hole detection to complete the detection of the installation holes on the installation station,obtain the spatial coordinate information.Fit the center point coordinates of the installation station by the least square method,that is,"centering";The center point of the position forms the fitting central axis,which serves as a guide for the advancement of the transportation and installation vehicle,that is,the "fixed axis".Design the centering error parameter and the fixed axis error parameter as the index to measure the positioning accuracy of the centering-fixing axis algorithm.The simulation verification results show that the centering-fixing axis algorithm based on binocular stereo vision meets the centering demand and the fixed axis demand,and can realize the accurate and fast positioning of the transportation and installation vehicle during the automatic installation of aero-engines. |
PDF Full Text Request